REF: simplify maybe_casted_values (pandas-dev#38421)

Author: jbrockmendel

pandas/core/dtypes/cast.py

@@ -88,17 +88,11 @@
     ABCSeries,
 )
 from pandas.core.dtypes.inference import is_list_like
-from pandas.core.dtypes.missing import (
-    is_valid_nat_for_dtype,
-    isna,
-    na_value_for_dtype,
-    notna,
-)
+from pandas.core.dtypes.missing import is_valid_nat_for_dtype, isna, notna
 
 if TYPE_CHECKING:
     from pandas import Series
     from pandas.core.arrays import ExtensionArray
-    from pandas.core.indexes.base import Index
 
 _int8_max = np.iinfo(np.int8).max
 _int16_max = np.iinfo(np.int16).max
@@ -488,53 +482,6 @@ def changeit():
     return result, False
 
 
-def maybe_casted_values(
-    index: "Index", codes: Optional[np.ndarray] = None
-) -> ArrayLike:
-    """
-    Convert an index, given directly or as a pair (level, code), to a 1D array.
-
-    Parameters
-    ----------
-    index : Index
-    codes : np.ndarray[intp] or None, default None
-
-    Returns
-    -------
-    ExtensionArray or ndarray
-        If codes is `None`, the values of `index`.
-        If codes is passed, an array obtained by taking from `index` the indices
-        contained in `codes`.
-    """
-
-    values = index._values
-    if values.dtype == np.object_:
-        values = lib.maybe_convert_objects(values)
-
-    # if we have the codes, extract the values with a mask
-    if codes is not None:
-        mask: np.ndarray = codes == -1
-
-        if mask.size > 0 and mask.all():
-            # we can have situations where the whole mask is -1,
-            # meaning there is nothing found in codes, so make all nan's
-
-            dtype = index.dtype
-            fill_value = na_value_for_dtype(dtype)
-            values = construct_1d_arraylike_from_scalar(fill_value, len(mask), dtype)
-
-        else:
-            values = values.take(codes)
-
-            if mask.any():
-                if isinstance(values, np.ndarray):
-                    values, _ = maybe_upcast_putmask(values, mask, np.nan)
-                else:
-                    values[mask] = np.nan
-
-    return values
-
-
 def maybe_promote(dtype, fill_value=np.nan):
     """
     Find the minimal dtype that can hold both the given dtype and fill_value.

pandas/core/frame.py

@@ -83,7 +83,6 @@
     infer_dtype_from_scalar,
     invalidate_string_dtypes,
     maybe_box_datetimelike,
-    maybe_casted_values,
     maybe_convert_platform,
     maybe_downcast_to_dtype,
     maybe_infer_to_datetimelike,
@@ -5021,8 +5020,18 @@ class    max    type
                     missing = self.columns.nlevels - len(name_lst)
                     name_lst += [col_fill] * missing
                     name = tuple(name_lst)
+
                 # to ndarray and maybe infer different dtype
-                level_values = maybe_casted_values(lev, lab)
+                level_values = lev._values
+                if level_values.dtype == np.object_:
+                    level_values = lib.maybe_convert_objects(level_values)
+
+                if lab is not None:
+                    # if we have the codes, extract the values with a mask
+                    level_values = algorithms.take(
+                        level_values, lab, allow_fill=True, fill_value=lev._na_value
+                    )
+
                 new_obj.insert(0, name, level_values)
 
         new_obj.index = new_index